I would like to know if there is any supplementary acados documentation.
I am currently using the Matlab interface and I would like to know if there is anywhere the explanation for what all input and output parameters mean and what their possible values are (i.e. gnsf_detect_structure, regularize method in input and res_g, res_b, res_d, res_m in output) . Many of them are in common with Acado (you can find the documentation here). However, the list here is not exhaustive and the values are sometimes different. Any help?
I guess a detailed documentation will be available soon, but the following papers might help you.
gnsf_detect_structure, roughly speaking it is about detecting linear structures in the dynamics to speed up the computations, see:
Further, the regularization method is about the Hessian in the QP subproblems in an SQP method. Depending on the QP solver, the full space (or reduced) Hessian has to be positive definite, since this is not always the case regularization of the Hessian is needed, for a overview what project,project_reduc_hess, mirror or convexify actually do and theoretical background see:
I hope this helps a bit for understanding the underlying methods.
I attach here the list of parameters of the acados’ matlab interface, with their meaning and a few references. This is not exhaustive, it comes from the pieces of information that I could put together, but it could be useful for new users. Unfortunately, editable formats such as .docx are not accepted on the post. I’ll share it in .jpg then.
Hi Enrica,
Thanks a lot, this is very useful.
I think we can add this cheatsheet style of documentation also on docs.acados.org.
In that case I suggest we create a Markdown or Latex version of it.
For now you can maybe upload the *.docx on some platform like google docs and share the link to it.
Best,
Tommaso
I have converted it to Markdown and created a pull request to include it in the acados documentation. In the description of the commit you can find the link to the shared file, it is a Google doc on Drive.
The best workflow to generate the table that I found for now is the following:
edit the shared file
download it as “param_acados_matlab.docx” and convert
it to Github-flavored markdown with pandoc with the following commad
